<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0"><channel><title>PyVideo.org - Giannis Ashiotis</title><link>https://pyvideo.org/</link><description></description><lastBuildDate>Wed, 22 Oct 2014 00:00:00 +0000</lastBuildDate><item><title>PyFAI: a Python library for high performance azimuthal integration on GPU</title><link>https://pyvideo.org/euroscipy-2014/pyfai-a-python-library-for-high-performance-azim.html</link><description>&lt;h3&gt;Description&lt;/h3&gt;&lt;p&gt;In the field of X-Ray diffraction, 2D area detectors like ccd or pixel
detectors have become popular in the last 15 years for diffraction
experiments thanks to the large solid-angle coverage and to their better
signal linearity. These detectors have a large sensitive area of
millions of pixels with high spatial resolution and are getting fast:
one kilo-Hertz is expected this year. The software package pyFAI we
present here has been designed to reduce X-ray 2D-diffraction images
into 1D curves (azimuthal integration) usable by other software for
in-depth analysis such as Rietveld refinement, ... Other averaging
patterns like 2D integration, image distortion, ... are also available.
PyFAI is a library featuring a clean pythonic interface and aiming at
being integrated into other software. But it also needs to cope with the
data deluge coming from the detector...&lt;/p&gt;
&lt;p&gt;In this contribution, we would like to highlight the performance reached
by this library. To get today’s computers best performances, one needs
to have parallelized code and azimuthal integration is not directly
parallelizable. After a scatter-to-gather transformation of the
algorithm, it got parallel (Cython implementation). Other optimizations
have been used to get the best performances out of GPU (compensated
summation, partial parallel reductions, ...).&lt;/p&gt;
</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Giannis Ashiotis</dc:creator><pubDate>Wed, 22 Oct 2014 00:00:00 +0000</pubDate><guid isPermaLink="false">tag:pyvideo.org,2014-10-22:/euroscipy-2014/pyfai-a-python-library-for-high-performance-azim.html</guid><category>EuroSciPy 2014</category></item></channel></rss>